Underground pipeline slurry leakage data acquisition method

Abstract

The invention discloses an underground pipeline slurry leakage data acquisition method, and relates to the technical field of pipeline slurry leakage acquisition, and the method comprises the following steps: building an acquisition system: building the data acquisition system by using a pipeline, a flange joint, a water pump, a water tank, a control valve, an infrared camera, a POE switch, a temperature control switch, a heating rod, an acquisition end, a distribution box and a matched air switch; and simulating a slurry leakage scene. According to the invention, the slurry leakage scene can be accurately simulated, and the acquired data can truly reflect the underground condition. According to the method, multi-angle collection is carried out by an infrared camera, multi-source data is obtained by combining an Internet of Things sensor, an equipment log and manual input, and then high-quality and complete data with outstanding key features can be obtained through cleaning, denoising and feature extraction, so that a solid foundation is provided for subsequent analysis and model training.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline leakage data collection, and in particular to a method for collecting underground pipeline leakage data. Background Art

[0002] In industrial production, underground pipeline transportation of slurry is a common process, widely used in industries such as mining, petrochemicals, and construction. For example, in mining, mined slurry is transported to the surface for subsequent processing via underground pipelines. In the petrochemical industry, pipelines are used to transport various chemical slurries to complete the production process. However, pipeline leakage is a frequent problem, resulting in numerous negative impacts on production. Leakages lead to slurry waste and increase production costs. Statistics show that some mining companies lose hundreds of thousands of yuan or more in slurry annually due to pipeline leakage. Furthermore, slurry leakage can corrode and damage underground equipment, shortening its lifespan and increasing maintenance costs and downtime. For example, a petrochemical company experienced corrosion of critical equipment due to pipeline leakage, resulting in repair costs reaching millions of yuan, and production losses exceeding tens of millions of yuan during the maintenance period. Furthermore, slurry leakage poses safety risks. Leaked slurry can contaminate the underground environment, affect the health of workers, and even cause accidents such as slips and poisoning. Moreover, if slurry leakage occurs near electrical equipment, it may cause electrical faults such as short circuits, threatening the safety of underground operations.

[0003] Currently, relying on intelligent video recognition technology to address pipeline leakage is an effective method. Real-time analysis of pipeline images can quickly detect leaks and issue timely alarms, allowing personnel to take appropriate measures. However, the actual production environment significantly limits the collection of pipeline leakage data. The underground environment is complex, with harsh conditions such as high temperature, high humidity, and strong electromagnetic interference. This makes it difficult for traditional data acquisition equipment to operate stably, and the accuracy and integrity of the collected data cannot be guaranteed. For example, ordinary cameras are prone to fogging in the high humidity environment underground, resulting in blurred images and an inability to accurately identify leaks. At the same time, strong electromagnetic interference can affect the normal operation of sensors, causing deviations in the collected equipment operating parameters.

[0004] Furthermore, the cramped underground space and complex pipeline distribution make it difficult to install data acquisition equipment at all potential leak locations, resulting in blind spots in data collection. Furthermore, the requirement for continuous production makes data collection challenging without disrupting normal operations. This results in insufficient data on pipeline leakage, limiting its diversity and representativeness. These issues complicate the development of pipeline leakage identification algorithms, hindering the establishment of efficient and accurate identification models and severely restricting the development and application of related technologies.

[0005] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0006] In view of the problems in the related art, the present invention proposes a method for collecting slurry leakage data of underground pipelines to overcome the above technical problems existing in the existing related art.

[0007] The technical solution of the present invention is achieved as follows:

[0008] A method for collecting slurry leakage data of underground pipelines, comprising the following steps:

[0009] Step S1, building a data acquisition system: using pipes, flange joints, water pumps, water tanks, control valves, infrared cameras, POE switches, temperature control switches, heating rods, acquisition terminals, distribution boxes and their supporting air switches to build a data acquisition system;

[0010] Step S2, simulating a slurry leakage scenario: heating the water in the water tank with a heating rod, controlling the water temperature with a temperature control switch, starting the water pump to supply water, adjusting the pipeline pressure by controlling the valve, and after turning on the water pump, adjusting the tightness of the flange head screws and the tension of the valve to simulate dripping, leakage, and spraying of the pipeline and elbow;

[0011] Step S3, data collection: using an infrared camera in conjunction with a collection terminal to perform all-round data collection on the simulated slurry leakage scene from multiple angles, and record image data of the slurry leakage;

[0012] Step S4, data collection and storage: Data related to underground pipeline leakage is collected based on IoT sensors, equipment logs, and manual input, including equipment operating parameters, leakage image data, environmental data, and timestamp information. Data is then sorted and stored in an orderly manner based on time series and equipment identification dimensions using a distributed file system and distributed database.

[0013] Step S5, data preprocessing: cleaning the collected data, removing outliers and duplicate data, using median filtering to denoise the noisy image data, and extracting key features related to slurry leakage from the original data;

[0014] Step S6, data analysis and mining: which includes at least performing correlation analysis to find the potential relationship between slurry leakage and equipment operating parameters and environmental factors; performing cluster analysis to classify the slurry leakage data; performing time series analysis to predict the development trend of slurry leakage using the autoregressive integrated moving average model (ARIMA);

[0015] Step S7, optimizing the algorithm model: for performing targeted optimization on the pipeline leakage identification algorithm model based on the results of data analysis and mining.

[0016] Furthermore, the data preprocessing includes: performing data cleaning, calibrating the data sequence to (x1, x2, ..., x n ), calculate the mean Expressed as:

[0017]

[0018] The calculated standard deviation σ is expressed as:

[0019]

[0020] Among them, if a data point x j satisfy It is then determined as an outlier and removed, and for duplicate data, it is identified and deleted by calculating the hash value of the data or comparing the unique identification field of the data.

[0021] Furthermore, the data preprocessing also includes: performing data denoising, including sorting all pixel values in a neighborhood window set around each pixel point in the image, taking the middle value as the new value of the pixel point, and the neighborhood window is a 3×3 or 5×5 window.

[0022] Furthermore, extracting key features related to slurry leakage from the original data includes the following steps:

[0023] For image data, the contrast of the gray-level co-occurrence matrix is calculated as the texture feature, including the following steps:

[0024] The calibration gray level co-occurrence matrix is P(i,j), and its contrast calculation formula is expressed as:

[0025]

[0026] Where N is the number of gray levels;

[0027] For equipment operating parameters and environmental data, statistical features and trend features are extracted.

[0028] Furthermore, the data analysis and mining includes: performing correlation analysis to determine potential correlation rules between slurry leakage and equipment operating parameters and environmental factors, including the following steps:

[0029] Calibration equipment operating parameter set X={x1,x2,x3,...,x m}, the leakage event is Y. Calculate the support (X∪Y) and confidence (X→Y) to determine the association rules, which are expressed as:

[0030]

[0031]

[0032] Among them, M is the total number of transactions, XY M is the number of transactions containing X and Y, X M is the number of transactions containing X.

[0033] Furthermore, the data analysis and mining includes: performing cluster analysis to classify the leakage data, including the following steps:

[0034] Calibration data set D = {x1, x2, ..., x n}, randomly select K initial cluster centers C={c1,c2,…,c K}. For each data point x i , calculate the Euclidean distance between it and each cluster center, expressed as:

[0035]

[0036] Among them, m is the dimension of the data, and x i Assigned to the cluster with the closest cluster center.

[0037] Furthermore, a time series analysis is conducted, using the autoregressive integrated moving average model (ARIMA) to analyze the leakage data and predict the development trend of the leakage, including the following steps:

[0038] Calibrate the stationary time series y t , the expression of the ARIMA model is:

[0039]

[0040] Where p is the autoregressive order, is the autoregressive coefficient, q is the moving average order, θ j is the moving average coefficient, ∈ t is a white noise sequence.

[0041] Furthermore, the optimization algorithm model includes the following steps:

[0042] Based on the features found by correlation analysis, they are added to the input of the pipeline leakage identification algorithm model to improve the model's ability to identify leakage;

[0043] Adjust the algorithm's classifier parameters based on the leakage type results obtained through cluster analysis;

[0044] Combined with the prediction results of time series analysis, multiple models are fused to improve the model's ability to identify slurry leakage.

[0045] Beneficial effects of the present invention:

[0046] 1. This invention simulates the underground environment to build a data collection system, accurately simulating slurry leakage scenarios and ensuring that the collected data truly reflects the underground conditions. Infrared cameras collect data from multiple angles, combined with IoT sensors, equipment logs, and manual data entry to acquire multi-source data. After cleaning, denoising, and feature extraction, high-quality, complete data with prominent key features is obtained, providing a solid foundation for subsequent analysis and model training.

[0047] 2. The present invention's correlation analysis identifies potential relationships between slurry leakage and equipment operating parameters and environmental factors, helping to clarify the cause of the leakage. For example, identifying specific temperature and pressure combinations with high leakage risk provides a basis for leak prevention. Cluster analysis categorizes leakage data, revealing characteristic patterns of different types of leakage, facilitating the development of targeted detection and treatment strategies. Time series analysis predicts leakage trends and provides early warning of leaks, such as predicting the scale and probability of leakage, allowing companies to prepare maintenance resources in advance and reduce losses.

[0048] 3. This invention optimizes the leakage identification model. Based on the results of correlation analysis, it adds feature dimensions, such as temperature change rate and pressure fluctuation, to the pipeline leakage identification algorithm model to enhance the model's ability to identify leakage. Based on cluster analysis results, it adjusts model parameters, such as the number of hidden layer nodes and learning rate, to enhance the model's ability to fit different leakage types. By integrating time series analysis prediction results with the leakage identification model, it improves identification sensitivity during predicted leakage periods, enabling more accurate detection of leakage events and enhancing the model's practicality and accuracy.

[0049] 4. By timely and accurately detecting slurry leakage, this invention can avoid slurry waste, equipment damage, environmental pollution, and safety accidents, ensuring the safety of underground workers. It also reduces equipment maintenance and downtime, improves production efficiency, reduces production costs, enhances the economic benefits and competitiveness of enterprises, and promotes the sustainable development of industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 1 is a flow chart of a method for collecting slurry leakage data in a downhole pipeline according to an embodiment of the present invention;

[0052] Figure 2 The figure is a schematic diagram of a scenario of a method for collecting slurry leakage data in a downhole pipeline according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0054] According to an embodiment of the present invention, a method for collecting slurry leakage data in a downhole pipeline is provided.

[0055] like Figure 1 As shown, the method for collecting slurry leakage data of a downhole pipeline according to an embodiment of the present invention includes the following steps:

[0056] Step S1, pre-build the data acquisition system, in a laboratory or a site simulating the underground environment, connect the pipelines through flange joints 2 to form a simulated pipeline line, and connect the water tank 4, water pump 3, and control valve 5 to the pipeline in sequence to ensure that the water can circulate in the pipeline, and at the same time install an infrared camera 6 at a preset position of the pipeline to capture the leakage image. The camera is connected to the acquisition terminal 9 through a POE switch 7 to realize the transmission of image data, and a heating rod 8 is installed in the water tank 4, and a temperature control switch is connected to the heating rod 8 and the distribution box 10. The circuit is controlled by the matching air switch of the distribution box 10 to achieve precise control of the water temperature.

[0057] Steps S2-S3 simulate slurry leakage and collect data. This involves activating heater rod 8 and using a temperature control switch to adjust the water temperature in water tank 4 to a temperature close to the actual downhole production environment. Water pump 3 is activated, and valve 5 is used to adjust the pipeline pressure to a set value. The tightness of the flange screws and the valve tension are gradually adjusted to simulate slurry leakage in the pipeline and elbows, such as dripping, leaking, or spraying. During the simulation, infrared camera 6 captures the slurry leakage scene from different angles, and data acquisition terminal 9 records the captured image data in real time.

[0058] Step S4 involves data collection and storage. During data collection, IoT sensors collect device operating parameters, including pipeline pressure, water flow, and water temperature. Related information is obtained from device logs, and ambient humidity and pH are manually entered. This data is categorized by time series and device ID, and stored using the distributed file system HDFS and the distributed database Cassandra.

[0059] Specifically, using time as the key, the leakage data related to a specific device collected every day is stored in a file block to facilitate subsequent query and analysis.

[0060] Step S5, pre-processing the collected data, includes the following steps:

[0061] Perform data cleaning and calibrate the data sequence to (x1, x2, ..., x n ), calculate the mean Expressed as:

[0062]

[0063] The calculated standard deviation σ is expressed as:

[0064]

[0065] Among them, if a data point x j satisfy It is then identified as an outlier and removed. For duplicate data, it is identified and deleted by calculating the hash value of the data or comparing the unique identification field of the data.

[0066] Data denoising is performed using a median filter on the collected noisy image data. Taking a 3×3 neighborhood window as an example, for each pixel in the image, the values of all pixels within its 3×3 neighborhood are sorted, and the median value is taken as the new value for that pixel. This process is repeated for each pixel in the image, effectively removing salt and pepper noise and improving image quality.

[0067] Perform feature extraction to extract key features from the preprocessed data.

[0068] For image data, the contrast of the gray-level co-occurrence matrix is calculated as the texture feature, including the following steps:

[0069] The calibration gray level co-occurrence matrix is P(i,j), and its contrast calculation formula is expressed as:

[0070]

[0071] Where N is the number of gray levels.

[0072] Among them, for equipment operating parameters and environmental data, statistical features are extracted, including: mean, maximum, minimum, change rate and trend characteristics.

[0073] Step S6, performing data analysis and mining, includes the following steps:

[0074] Conduct correlation analysis to determine potential correlation rules between slurry leakage and equipment operating parameters and environmental factors. This includes the following steps:

[0075] Calibration equipment operating parameter set X={x1,x2,x3,...,x m}, the leakage event is Y. Calculate the support (X∪Y) and confidence (X→Y) to determine the association rules, which are expressed as:

[0076]

[0077] Among them, M is the total number of transactions, XY M is the number of transactions containing X and Y, X M is the number of transactions containing X.

[0078] In this technical solution, if it is found that when temperature and pressure meet certain conditions at the same time, the support and confidence of slurry leakage are high, it can be considered that these two factors are related to the slurry leakage.

[0079] Cluster analysis is performed to classify leakage data, including the following steps:

[0080] Calibration data set D = {x1, x2, ..., x n}, randomly select K initial cluster centers C={c1,c2,…,c K}. For each data point x i , calculate the Euclidean distance between it and each cluster center, expressed as:

[0081]

[0082] Among them, m is the dimension of the data, and x i The data is assigned to the cluster with the closest cluster center. The cluster center of each cluster is then recalculated, and the above process is repeated until the cluster center remains unchanged or changes very little. Cluster analysis can reveal characteristic patterns of different types of slurry leakage. For example, some slurry leakage data exhibits similarities in characteristics such as pressure and flow, and can be grouped together.

[0083] Perform time series analysis on the leakage data using the autoregressive integrated moving average model (ARIMA) to predict the development trend of leakage, including the following steps:

[0084] Calibrate the stationary time series y t , the expression of the ARIMA model is:

[0085]

[0086] Where p is the autoregressive order, is the autoregressive coefficient, q is the moving average order, θ j is the moving average coefficient, ∈ t is a white noise sequence.

[0087] This technical solution uses historical leakage data to fit and estimate parameters to determine the optimal model parameters, thereby predicting the development trend of leakage. For example, it can predict the probability and scale of leakage in the future.

[0088] Step S7, performing an optimization algorithm model, for performing targeted optimization on the pipeline leakage identification algorithm model based on the results of data analysis and mining, includes the following steps:

[0089] Based on correlation analysis optimization, if correlation analysis reveals specific temperature and pressure change trends before a slurry leak occurs, such as an increased likelihood of a slurry leak when the temperature rises and the pressure fluctuations increase, machine learning models such as decision trees and support vector machines can be retrained using the temperature change rate and pressure fluctuation values as new feature dimensions to improve their ability to identify slurry leaks.

[0090] Based on cluster analysis optimization, and based on the leakage type results obtained through cluster analysis, using a neural network model as an example, parameters such as the number of hidden layer nodes, learning rate, and weight initialization method are adjusted according to the characteristics of different leakage types. For complex leakage types, the number of hidden layer nodes is increased to improve the model's fitting ability. The optimal learning rate is determined through methods such as cross-validation, allowing the model to converge to the optimal solution more quickly.

[0091] Based on time series analysis optimization and combined with the forecast results of time series analysis, a machine learning-based slurry leakage identification model is integrated with the ARIMA forecasting model. In application, the sensitivity of the identification model is increased for time periods when slurry leakage is predicted, thereby more accurately detecting slurry leakage events. For example, when the ARIMA model predicts the possibility of slurry leakage in a certain period in the future, the threshold of the machine learning model is adjusted during that period to make it easier to identify slurry leakage.

[0092] In summary, with the help of the above technical solution of the present invention, the following effects can be achieved:

[0093] 1. This invention simulates the underground environment to build a data collection system, accurately simulating slurry leakage scenarios and ensuring that the collected data truly reflects the underground conditions. Six infrared cameras collect data from multiple angles, combining IoT sensors, equipment logs, and manual data entry to acquire multi-source data. After cleaning, denoising, and feature extraction, high-quality, complete data with prominent key features is obtained, providing a solid foundation for subsequent analysis and model training.

[0094] 2. The present invention's correlation analysis identifies potential relationships between slurry leakage and equipment operating parameters and environmental factors, helping to clarify the cause of the leakage. For example, identifying specific temperature and pressure combinations with high leakage risk provides a basis for leak prevention. Cluster analysis categorizes leakage data, revealing characteristic patterns of different types of leakage, facilitating the development of targeted detection and treatment strategies. Time series analysis predicts leakage trends and provides early warning of leaks, such as predicting the scale and probability of leakage, allowing companies to prepare maintenance resources in advance and reduce losses.

[0095] 3. This invention optimizes the leakage identification model. Based on the results of correlation analysis, it adds feature dimensions, such as temperature change rate and pressure fluctuation, to the pipeline leakage identification algorithm model to enhance the model's ability to identify leakage. Based on cluster analysis results, it adjusts model parameters, such as the number of hidden layer nodes and learning rate, to enhance the model's ability to fit different leakage types. By integrating time series analysis prediction results with the leakage identification model, it improves identification sensitivity during predicted leakage periods, enabling more accurate detection of leakage events and enhancing the model's practicality and accuracy.

[0096] 4. By timely and accurately detecting slurry leakage, this invention can avoid slurry waste, equipment damage, environmental pollution, and safety accidents, ensuring the safety of underground workers. It also reduces equipment maintenance and downtime, improves production efficiency, reduces production costs, enhances the economic benefits and competitiveness of enterprises, and promotes the sustainable development of industrial production.

[0097] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily conceive of other embodiments of the present invention after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed in the present invention. The description and examples are to be considered merely exemplary, and the true scope and spirit of the present invention are indicated by the claims.

[0098] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for collecting underground pipeline leakage data, characterized in that: The following steps are involved: Step S1, building a data acquisition system: using a pipeline (1), a flange joint (2), a water pump (3), a water tank (4), a control valve (5), an infrared camera (6), a POE switch (7), a temperature control switch, a heating rod (8), a data acquisition terminal (9), a distribution box (10) and its supporting air switch to build a data acquisition system; Step S2, simulating a slurry leakage scenario: heating the water in the water tank (4) by means of a heating rod (8), controlling the water temperature by means of a temperature control switch, starting the water pump (3) to supply water, regulating the pipeline pressure by means of a control valve (5), and after turning on the water pump (3), adjusting the tightness of the flange head screws and the tension of the valve to simulate dripping, leakage, and spraying of liquid in the pipeline and elbow; Step S3, data collection: using the infrared camera (6) in conjunction with the collection terminal (9), to collect all-round data of the simulated slurry leakage scene from multiple angles and record the image data of the slurry leakage; Step S4, data collection and storage: Data related to underground pipeline leakage is collected based on IoT sensors, equipment logs, and manual input, including equipment operating parameters, leakage image data, environmental data, and timestamp information. Data is then sorted and stored in an orderly manner based on time series and equipment identification dimensions using a distributed file system and distributed database. Step S5, data preprocessing: cleaning the collected data, removing outliers and duplicate data, using median filtering to denoise the noisy image data, and extracting key features related to slurry leakage from the original data; Step S6, data analysis and mining: which includes at least performing correlation analysis to find the potential relationship between slurry leakage and equipment operating parameters and environmental factors; performing cluster analysis to classify the slurry leakage data; performing time series analysis to predict the development trend of slurry leakage using the autoregressive integrated moving average model (ARIMA); Step S7, optimizing the algorithm model: for performing targeted optimization on the pipeline leakage identification algorithm model based on the results of data analysis and mining.

2. The method for collecting underground pipeline leakage data according to claim 1, characterized in that: The data preprocessing includes: performing data cleaning, calibrating the data sequence to (x1, x2, ..., x n ), calculate the mean Expressed as: The calculated standard deviation σ is expressed as: Among them, if a data point x j satisfy It is then determined as an outlier and removed, and for duplicate data, it is identified and deleted by calculating the hash value of the data or comparing the unique identification field of the data.

3. The method for collecting underground pipeline leakage data according to claim 2, characterized in that: The data preprocessing also includes: performing data denoising, including sorting all pixel values in a neighborhood window set around each pixel point in the image, taking the middle value as the new value of the pixel point, and the neighborhood window is a 3×3 or 5×5 window.

4. The method for collecting underground pipeline leakage data according to claim 3, characterized in that: The method of extracting key features related to slurry leakage from the original data includes the following steps: For image data, the contrast of the gray-level co-occurrence matrix is calculated as the texture feature, including the following steps: The calibration gray level co-occurrence matrix is P(i,j), and its contrast calculation formula is expressed as: Where N is the number of gray levels; For equipment operating parameters and environmental data, statistical features and trend features are extracted.

5. The underground pipeline leakage data acquisition method according to claim 1, characterized in that: The data analysis and mining includes: performing association analysis to determine potential association rules between slurry leakage and equipment operating parameters and environmental factors, including the following steps: Calibration equipment operating parameter set X={x1,x2,x3,...,x m }, the leakage event is Y; calculate the support (X∪Y) and confidence (X→Y) to determine the association rules, which are expressed as: Among them, M is the total number of transactions, XY M is the number of transactions containing X and Y, X M is the number of transactions containing X.

6. The method for collecting underground pipeline leakage data according to claim 5, characterized in that: The data analysis and mining includes: performing cluster analysis to classify the leakage data, including the following steps: Calibration data set D = {x1, x2, ..., x n }, randomly select K initial cluster centers C={c1,c2,…,c K }; For each data point x i , calculate the Euclidean distance between it and each cluster center, expressed as: Among them, m is the dimension of the data, and x i Assigned to the cluster with the closest cluster center.

7. The method for collecting underground pipeline leakage data according to claim 6, characterized in that: Perform time series analysis on the leakage data using the autoregressive integrated moving average model (ARIMA) to predict the development trend of leakage, including the following steps: Calibrate the stationary time series y t , the expression of the ARIMA model is: Where p is the autoregressive order, is the autoregressive coefficient, q is the moving average order, θ j is the moving average coefficient, ∈ t is a white noise sequence.

8. The method for collecting underground pipeline leakage data according to claim 1, characterized in that: The optimization algorithm model includes the following steps: Based on the features found by correlation analysis, they are added to the input of the pipeline leakage identification algorithm model to improve the model's ability to identify leakage; Adjust the algorithm's classifier parameters based on the leakage type results obtained through cluster analysis; Combined with the prediction results of time series analysis, multiple models are fused to improve the model's ability to identify slurry leakage.

Images